By combining incidence and mortality rates recorded for the entire population with relative risk measures and exposure prevalence estimates derived from case-control and cohort studies, we estimate the likelihood of developing cancer, in the next five years or over the lifetime, according the the established risk factors. We are examining how these risk prediction models are influenced by uncertainty in the data, especially at the range of risks that may lead to clinical decisions for screening, surgery, or chemoprevention.